A Partitioning Technique for a Waveform Relaxation Method Using Eigenvectors in the Transient Stability Analysis of Power Systems

It is of paramount importance that power system operators be able to assess transient stability in order to realize a reliable and stable power supply. Transient stability analysis can be formulated as a large-scale system of differential and algebraic equations (DAE). However, as power systems are becoming larger and more complex, it is becoming difficult to solve DAE in a practical amount of time for system operations. Parallel computing based on the waveform relaxation method is an effective solution to achieve faster calculations for transient stability analysis. To enhance the performance of the waveform relaxation method, a proper partitioning of the original problem is essential. Although various partitioning approaches have been used, those approaches might not be effective when analyzing a weakly damped low-frequency oscillation. In particular, in the Japanese 60-Hz power system, this oscillation becomes an important problem. To resolve this issue, in this paper we have developed a new partitioning method that is better suited to analyzing a weakly damped low-frequency oscillation based on eigenvalue analysis. Specifically, effective partitioning can be automatically determined by the proposed index, which can evaluate the validity of the partitioning. The proposed method was tested using the Japanese standards of the IEEJ WEST10 system model and the WEST30 system model.